Meta Launches Muse Wonder: Pricing, Access and Data Terms B2B Teams Should Check

Meta announced the models on Monday, October 12, 2026, according to TechCrunch and VentureBeat. TechCrunch describes them as Meta's first attempt at a frontier model, larger than the lightweight Muse Spark that Meta built for low latency. Both models will also power the Muse agent.

For B2B software buyers, the interesting parts are the price, the access split between the two models, and the data handling terms. Benchmark claims come mostly from Meta's own model card, so treat them as vendor-reported until independent testing accumulates.

What Meta Released and When

Meta released Muse Wonder and Muse Wonder Full on October 12, 2026. Muse Wonder is available through the Meta Model API, while Muse Wonder Full is not served through the API and is limited to Muse Code subscribers.

• Both models are proprietary, with no publicly downloadable weights, per VentureBeat.
• VentureBeat reports text, image, audio and video input, text output and a context window of about 1 million tokens.
• The API identifier is muse-wonder-1.0, with rate limits of 3,000 requests and 4 million tokens per minute.

Pricing and Plans

The pay-as-you-go API charges $4 per million input tokens and $20 per million output tokens for Muse Wonder, with cached input at $0.20 per million tokens, according to VentureBeat. TechCrunch calls that roughly equivalent in cost to GPT-6 Sol or Claude Opus 5.5.

• Web search through the API adds $2.50 per 1,000 queries.
• Muse Code Power Usage costs $50 per month and Maximum Usage costs $200 per month, which unlock Wonder Full.
• Artificial Analysis estimates an average cost per task of $4.52 for Wonder, $5.98 for Opus 5.5 and $3.26 for GPT-6 Astra, as cited by VentureBeat.

Performance Claims and Caveats

Meta says both models score comparably to leading frontier systems, and VentureBeat cites an Artificial Analysis Intelligence Index score of 56 for Wonder Full and 52 for standard Wonder. VentureBeat also notes that Meta's comparison table mixes third-party and internal evaluations.

That means the numbers are not fully controlled comparisons. Run your own evaluation on your own tasks before changing any production model choice.

Why It Matters for B2B SaaS Buyers

A new frontier-class option from a large vendor adds pricing pressure and another supplier to evaluate, but it does not remove the need for governance. Data handling terms and the split between API and subscription access matter as much as raw capability.

• VentureBeat reports that the API runs in a mode where submissions are not used for training, while Muse Code subscriptions default to a contributor mode that allows Meta to use submitted content to improve future models.
• Zero data retention is available to qualified API organizations through separate approval and excludes some features such as file uploads and built-in web search.
• Because Wonder Full is not on the API, you cannot call it from your own SaaS products or workflows.

What IT Teams Should Do Now

Treat the launch as an evaluation prompt, not an urgent migration. Update your approved-model list only after security review, a cost test on real workloads and a check that contract and data terms match your policy.

• Check whether any team is using Muse Code under the default contributor mode with company code or confidential data.
• Benchmark Muse Wonder against your current model on three to five real tasks and compare cost per completed task, not token price alone.
• Review your AI vendor risk process against the NIST AI Risk Management Framework.
• Confirm rate limits, data residency and retention terms before building customer-facing features.
• Keep a second model provider in place to avoid lock-in.

Related Reading on SaaSPodium

Teams that build or buy AI features usually need several adjacent tools around the model itself, from code assistants to gateways that govern access. If you are tracking AI tooling for your stack, browse the AI code generators category and the AI writing assistant category, and compare API management software for governing model access.

Vendor details are changing quickly, so confirm current pricing and terms on Meta AI, as this article reflects reporting from October 12, 2026.

Frequently Asked Questions

How much does the Meta Muse Wonder API cost?
Per VentureBeat, Muse Wonder costs $4 per million input tokens and $20 per million output tokens, with cached input at $0.20 per million tokens and web search at $2.50 per 1,000 queries. Confirm current rates in Meta's official documentation before budgeting.

Can companies use Muse Wonder Full through the API?
No. TechCrunch and VentureBeat both report that Muse Wonder Full is not served through the API and is limited to Muse Code subscribers. Only the standard Muse Wonder model is available through the Meta Model API.

Does Meta train on data sent to Muse Wonder?
VentureBeat reports that the pay-as-you-go API runs in a standard mode where submissions are not used for training, while Muse Code subscriptions default to a contributor mode that allows it. Review Meta's terms and your data policy before sending sensitive content.